Chromantis

Probabilistic nowcasting and four-week forecasting of influenza, COVID-19 and RSV admissions and ED visits. Independently fitted stages share one model interface and connect through sampled reconstructed histories.

• Documentation
• Architecture and reflection
• Workflow: acquire, build, plan, launch and rank
• Model runs: dated forecasting reports; nowcasting section
• Data and interactive staircase
• Setup

uv sync
.venv/bin/python -m chromantis.dataset.build show
.venv/bin/python -m chromantis.experiment.planner status -e b-2-t0

planner rank is the canonical analysis path. Reports live in docs/experiments/; model artifacts live in data/experiments/. The two-stage nowcasting/forecasting pipeline remains available through task=pipeline; standalone tasks use task=nowcast or task=forecast. See the workflow for full manager commands.